Navigating the Age of Interdependency: Why a New Approach to Application Performance Management is Needed

Websites and Applications are Growing More Intertwined Every Day

Last summer, Facebook (News - Alert) suffered a period of sporadic outages. All around the Web, thousands of other sites with a Facebook “Like” button slowed to a crawl. About a month later, Amazon’s EC2 outage took several big sites offline and slowed down many others. In the fall, a technical malfunction at GoDaddy took down its website and millions more; while a Google (News - Alert) App Engine disruption affected thousands of customers around the world that use the platform-as-a-service.

A few short years ago, an outage on a prominent website would have resulted in only that single site being impacted. But because today’s modern applications call services from a variety of third-party providers, outages now cascade throughout the Internet and bring other applications and services down with them. The recent issues with Facebook, Amazon, GoDaddy and Google App Engine are just the latest examples of this growing interdependency and the downside that can result from integrating with third-party services for functionalities like advertising, social plug-ins, video, newsfeeds and many others. Compuware’s (News - Alert) latest research shows the typical U.S. website can have more than a dozen separate hosts contributing to a single transaction, and this interdependency will only continue to grow.

Focusing Where it Counts: The True End-User Experience

Today, serving a website or application now means making sure all these pieces assemble in a way that yields the best possible service to the end user. From an IT perspective, this means it’s no longer enough to solely control the factors within your firewall; you must also mitigate the risks imposed by third-party services that can slow your website’s performance or take it down. This is the only way for today’s IT executives to meet the challenge of an increasingly interconnected Web.

This of course raises a critical question: how can I manage my applications performance when it involves so many external elements out of my direct control? Finding the answer requires a new approach to application performance management (APM (News - Alert)) that centers less on the data center and more on the end user. In other words, businesses need to measure and monitor application performance from the true end-user perspective. Gartner (News - Alert) notes this sea change is already happening: it estimates that 20 percent of the Global 2000 are trying to reconstruct their IT operational process frameworks in ways that accord with the monitoring and management of applications, rather than infrastructure, in a central place.

Other Keys to Mitigating Third-Party Risks

As website and application interdependencies increase, organizations will initiate comprehensive processes to help ensure the successful implementation and continued functioning of third-party Web services. All centered on revealing the true end-user performance impact, these processes can be implemented in a time- and cost-efficient manner and include:

• Benchmarking response time and availability of each third-party web component before signing contracts.
• Designing pages so third-party providers’ scripts load asynchronously in parallel with the crucial page content.
• Testing components before launch in multiple phases and under various conditions. This will entail pre-production testing for a proof of concept and load testing to evaluate performance under conditions similar to what may be expected on a website and during high traffic periods.
• Devising fast-fail programs that secure the functioning of an overall website, even if any one particular component should crash.
• Considering redundant services for the most critical features of a site, such as the shopping cart.

Once organizations have pre-tested third-party components to assess their potential impact on performance, the real work will begin. Even though third-party vendors may be contracted to maintain certain standards, organizations will ultimately increase their vigilance in ensuring these standards are met, by:  

• Setting measurable and objective levels for SLAs that reference accurate data all parties can see and understand. Without defined standards, there is no way to hold vendors to exact account.
• Measuring over time. The occasional snapshot won’t do it; to ensure quality under all loads and across all markets, organizations will increasingly implement a regime of 24/7 application monitoring.
• Automating the removal of a failing object (e.g. ads, tracking pixels) when it’s been identified as an issue, and the process of bringing it back once it’s working again.
• Sharing the wealth of data with one’s third-party vendors. The objective measurements an organization gathers allows it to work amicably with third-party partners to align interests and resolve problems.

Third-party services can be incredibly valuable in enriching your website’s ability to drive more traffic to your site, offer interactive experiences, increase conversions and add new functionality when it becomes available. But the peril is that these services can also make your site vulnerable to degraded performance or a complete shutdown if any one of these components fails.

Because websites and applications are the lifeblood of today’s businesses, you can’t afford any unexpected surprises in the form of performance drop-offs. It is more critical than ever for organizations to know how to mitigate third party performance issues. In the near-term, we expect many organizations to put the necessary steps and processes in place to address this new reality.